Real Temp - New temp program for Intel Core processors

[unclewebb]

Ace-a-Rue: I heard on another forum that someone is using RealTemp on Vista x64. Supposedly they swapped in the open source 64 bit drivers and it worked.

Want to do some x64 beta testing? I comibined the x64 drivers with my program. I'll have a smile from ear to ear if it's that simple to get it working in 64 bit. My wallet will thank the first person to post a x64 screen shot.

http://www.fileden.com/files/2008/3/...ealTemp_64.zip

[jaredpace]

There is an interesting submission at Anandtech.



More info here: http://anandtech.com/cpuchipsets/int...spx?i=3251&p=4

Is this really a sign that someone at Intel has been checking out this forum and discovered that we're way too close to the truth now so it was time for a PR to try and throw us off the trail! Maybe Intel isn't very happy with software that shows that their DTS sensors sometimes get stuck on their new 45nm chips. This wasn't an issue before with 65nm.

Why don't we just make it a standard to measure margins from here on out?

Like, "hey joe, don't let your e8400 get within 12C of Tjmax."

or is this impractical?

[unclewebb]

With my E6400 that was the conclusion I came to. To overclock this amount reliably, I need to leave myself 25C of headroom before TjMax. That was perfect and I thought it should apply to my new E8400 as well but it doesn't. Intel may have raised their TjMax spec 10C from 85C to 95C but when overclocking and over volting I am limited by the absolute core temperature of my processor. I'm finding that the absolute temperature limit of these 2 processors is almost identical which means with my E8400 I now need 35C of head room to run reliably. A new TjMax rating doesn't seem to have changed the safe, stable temperature that these processors can run Prime stable at.

[jaredpace]

With my E6400 that was the conclusion I came to. To overclock this amount reliably, I need to leave myself 25C of headroom before TjMax. That was perfect and I thought it should apply to my new E8400 as well but it doesn't. Intel may have raised their TjMax spec 10C from 85C to 95C but when overclocking and over volting I am limited by the absolute core temperature of my processor. I'm finding that the absolute temperature limit of these 2 processors is almost identical which means with my E8400 I now need 35C of head room to run reliably. A new TjMax rating doesn't seem to have changed the safe, stable temperature that these processors can run Prime stable at.

so speaking for e8400's:

if tjmax is 95C, and you need 35C headroom, that would make 60C your maximum safe operating temp when priming. As you've said.

[theonlybabyface]

Ace-a-Rue: I heard on another forum that someone is using RealTemp on Vista x64. Supposedly they swapped in the open source 64 bit drivers and it worked.

Want to do some x64 beta testing? I comibined the x64 drivers with my program. I'll have a smile from ear to ear if it's that simple to get it working in 64 bit. My wallet will thank the first person to post a x64 screen shot.

http://www.fileden.com/files/2008/3/...ealTemp_64.zip

Just tried it for you. No worky. Complains that WinRing0.dll is missing although it's there.

[Retrospekt]

So CoreTemp tells me the TjMax is 100c, but RealTemp says 85. Which should I trust?

If you guys say RealTemp..I'm going way higher with my OC.

[unclewebb]

so speaking for e8400's:

if tjmax is 95C, and you need 35C headroom, that would make 60C your maximum safe operating temp when priming. As you've said.

I can Prime reliably at higher temperatures but I have to reduce my maximum overclock to do it. That's sort of the relationship. The cooler you can run your processor, the more MHz you can run it reliably at. That's why people like things like liquid nitrogen.

theonlybabyface: Maybe I misread the other post or it was an error. You could create a new folder and try copying the regular version of Real Temp in with these different drivers and see if that works. I'll try to do some more research on this.

[unclewebb]

Retrospekt: Ignore both programs because when it comes to overclocking it doesn't really matter. Take it to the moon and then crank it up a little more.

When overclocking your processor, it will reboot while running Prime if you have gone too far. If you are Prime stable then you are good to go.

[[XC] 4X4N]

Just tried it for you. No worky. Complains that WinRing0.dll is missing although it's there.

I get the same thing. Here is a shot for you.

[scottc19]

I also got this message, it is fully confirmed

[Retrospekt]

Retrospekt: Ignore both programs because when it comes to overclocking it doesn't really matter. Take it to the moon and then crank it up a little more.

When overclocking your processor, it will reboot while running Prime if you have gone too far. If you are Prime stable then you are good to go.

That is the first time I have ever heard that. Everyone else says 60 tops, 65 for bench. O_o

[Infa]

but most of the people are so freakin scared, they think that the comp will blow if the use 1.50 etc...hahah...and making such a scene about it..

i agree with unclewebb completely. if it hold together and not trhottling under heavy load, then ýour good to goo!

and BTW, Intel have speced it to 71c i think so why stop at 60? seems so not a good idea

[unclewebb]

For my well tested E6400 this is what I found:

3200 MHz --- DTS = 5 --- 80C
3400 MHz --- DTS = 15 --- 70C
3600 MHz --- DTS = 25 --- 60C

It was a pretty clear relationship. At 3200 MHz my computer was Prime stable at an absolute core temperature of 80C. I could even go higher and have it start throttling but it wouldn't stop or crash. To get more MHz I needed more core voltage and I had to keep adding more temperature headroom to remain Prime stable. Maybe I'll throw my E6400 back in tomorrow for some high temp abuse!

Thanks for the guys who tried to make this run under x64. I knew that post I read sounded too good to be true. Now I'll have to find it again and ask the guy what his secret really is or get him to post a screen to prove it.

http://forums.anandtech.com/messagev...VIEWTMP=Single

[xgman]

It didn't work with phase for zero temps and said all 4 cores were stuck also. Asus boards juts dopn't read minus temps any more.

[Demo]

Q6600 G0 Stepping
1.335v bios voltage, 1.248 ~ 1.256v loaded with prime95 v25.6 SmallFTT test.
Real temp is reading about 5c lower, calibrated the readings using "0"

Is real temp showing me the correct temps ?

[RocKer]

If i read correct,it does'nt work for XPpro 64bit?

Any idea when it will work for 64bit.

[Ace-a-Rue]

unclewebb is working on it right now...BUT...he is under the weather with a heavy cold currently.

[unclewebb]

Is real temp showing me the correct temps ?

Here's the question people need to start asking.
Is CoreTemp showing me the correct temps?

[rusty]

God bless unclewebb and ixtapalapaquetl and whoever else is working on realtemp !

[unclewebb]

It didn't work with phase for zero temps and said all 4 cores were stuck also.

RealTemp is capable of reporting sub zero temperatures but only if the DTS sensors are capable of operating at sub zero temps.

DTS data is stored in 7 bits of data so the maximum it can hold is 127.

Reported Temps = TjMax - DTS
Reported Temps = 95 - 127 = -32C

Maybe Intel will release a line of truly extreme processors that have DTS sensors that remain functional down at these low temperatures.

For me it's time for a break from programming. A minimize to system tray feature is next weeks goal. I don't currently have a 64 bit OS to test on so x64 development is going to have to wait for a while.

God bless unclewebb and ixtapalapaquetl and whoever else is working on realtemp !

Thank-you Rusty! I had trouble sleeping last night because I was afraid that the secret police were going to show up and take me away because I was getting too close to the truth.

[fornowagain]

You could create a new folder and try copying the regular version of Real Temp in with these different drivers and see if that works. I'll try to do some more research on this.

Nope still doesn't work, Vista64.

[WoZZeR999]

I got Real temp running using the compiled winring0 from http://dl.crystaldew.info/download/WinRing0_1_1_2.zip.

I'm running x64. I don't know if the temps are right (they look to be). Test sensors shows 11-12.



EDIT: I guess instructions on what I did may help. I just opened the zip file, and extracted it all to the Realtemp folder.

EDIT 2: I did this using the Realtemp-x64 from above.

[Warboy]

Am I getting this right? I did what Wozzer did and used a different Winring0 to get it to work on Vista 64, but This doesn't seem right.

[WoZZeR999]

Have you run the test sensors on a 32-bit OS? I got 11-12 (I have only run it on x64), so I'm not sure if it's a problem with the newer winring, or the fact that it's a quad. I would assume that the API for the winring wouldn't have changed that much. Features are normally added or tweaked, but usually backwards compatible.

[fornowagain]

Having done low power low ambient tests before to make sure calculated casetemps are reasonably accurate.
-3 ambient, 3200GHz 1.2V (2GPM PA120.3+Fuzion)

Ambient + system load (inc pump) x radiator C/W + CPU Load x (block C/W + TIM C/W)

-3 + 0.017*(137+18) + 137*(0.057 + 0.0045) = -3 + 2.6 + 8.4 = 8C




--------------------------

Vista64 QX9650 - Ambient 20C. Loop temp 21.4C. Knowing my systems thermal resistances quite well. Gives an idle of 60W and a final case temp around 25C.

20 + 0.017*(60+18) + 60*(0.057 + 0.0045) = 20 + 1.35 + 3.7 = 25C


--------------

Full load of prime95 small ffts at 4GHz is around 240W. Room ambient is 20C after running prime till temps stabilise, loop is 24.4C

Ambient + system load (inc pump) x radiator C/W + CPU Load x (block C/W + TIM C/W)

20 + 0.017*(240+18) + 240*(0.057 + 0.0045) = 20 + 4.3 + 14.7 = 39C casetemp.

If the coretemp was say around 3C higher than casetemp. It then would then be spot on at '0'.
If the coretemp was say around 13C higher than casetemp. It then would then be spot on at '0' and the Tjmax would must be 105C.